Tunnels and Underground Cities: Engineering and Innovation meet Archaeology,
Architecture and Art – Peila, Viggiani & Celestino (Eds)
© 2019 Taylor & Francis Group, London, ISBN 978-1-138-38865-9
Fiber-reinforced concrete segments: Comparison of simplified
analyses and FE simulations for the 3RPORT CSO Project, Fort
Wayne, Indiana, United States
D. Chapman & J. Parkes
Schnabel Engineering, Morristown, New Jersey, USA
G. Venturini & R. Comini
SWS Engineering Spa, Trento, Italy
L. Waddell
Lane Construction Corporation, Fort Wayne, Indiana
ABSTRACT: The design of steel fiber reinforced concrete (FRC) precast segmental liners was
recently developed for the 3RPORT Project following American Concrete Institute (ACI) require-
ments and European standards. Nine load cases were analyzed including final ground and water
loads with up to 8 bar of external hydrostatic pressure, as well as fabrication, handling, and TBM
thrust loads. The TBM thrust produces concentrated forces that cause bursting and spalling stres-
ses. Simplified analysis methods cannot describe the development of spalling stresses between
TBM jack pads and do not account for the ductility of the steel fibers. Consequently, reinforce-
ment bars may be unnecessarily prescribed based on analyses using such methods. Advanced 3D
numerical models were developed including the use of a constitutive law for FRC. This allowed
evaluation of post-cracking behavior during TBM thrust, crack width progress and consequent
stress redistribution. These results justified exclusive use of steel fiber reinforcement.
1 INTRODUCTION
The Three Rivers Protection & Overflow Reduction Tunnel (3RPORT) is a combined sewer
overflow (CSO) tunnel project located in Fort Wayne, Indiana. The tunnel and the associated
pipe network are part of the City’s Long-Term Control Plan for reducing the amount of com-
bined sewage that is discharged into Fort Wayne’s rivers every year.
The tunnel will be constructed in rock at depths up to about 70m below grade using a pres-
surized-face (slurry) tunnel boring machine (TBM). The projected life for the 7,600m, 4.88m
internal diameter (ID) tunnel is 100 years.
The approximately $188 million project, when completed, will reduce combined sewage
overflows to the St. Mary’s and Maumee Rivers by 90 percent from over 70 per year to just
four. The project was awarded to the Salini-Impregilo/S.A. Healy Joint Venture, now Lane
construction (Lane), for construction in April, 2017.
2 TUNNEL LINING REQUIREMENTS
2.1 Specification Requirements
The tunnel lining will be constructed using bolted and gasketed precast concrete tunnel lining
segments. Schnabel-SWS was engaged by Lane and by their precast segment supplier, CSI-
Forterra Fort Wayne JV to perform the design of the precast segmental lining.
1917
The design was based on requirements in the project specifications, drawings, and external
loads provided in the GBR. The primary requirements pointed to a design using steel fiber
reinforcement in lieu of traditional steel reinforcing bars. Key design requirements included:
– Design to be based on ACI 350 (2006). Do not increase allowable stresses for temporary
loads, or reduce load factors to account for temporary structure design.
– Avoid cruciform joints and ensure that TBM jacking shoes do not cross radial joints.
– Minimum ID of 4.88m and minimum thickness of 305mm.
– Minimum 28-day concrete compressive strength of 55 MPa
– Tunnel lining to be designed using steel fiber, minimum dosage 21 kg/m3, with additional
requirements:
– Minimum flexural tensile strength at 28 days of 670 psi per ASTM C1609. This was
interpreted to be equivalent to the first peak strength, f1, (ASTM 1609).
– Minimum post-crack equivalent residual flexural tensile strength at any deflection at or
beyond the deflection of L/600 of 460 psi per ASTM C1609. This is interpreted to be
equivalent to the residual strength at or beyond a deflection of L/600 per ASTM C1609,
fD600, per paragraph 3.2.14 of ASTM C1609.
– Gaskets to be designed for a minimum working pressure of 8 bars above atmospheric with
additional performance requirements for any gap between adjacent segments, and for
150 percent of the maximum tail void grouting pressure specified.
– Bolted connections at radial joints; bolts or dowels for circumferential joints.
– Minimum concrete cover to steel reinforcement and minimum steel ratio of 0.5% of gross
concrete area radially and circumferentially specified for possible rebar alternative.
2.2 Segment Manufacturer Requirements
The segment ring consists of 6 segments as shown in Figure 1.
CSI-Forterra provided the following additional parameters of the planned segment design:
– Nominal segment ring width of 60 inches with a maximum ring taper of 1.25 inches, in
order to meet the curve radii of the geometric design of the tunnel route.
– Basic ring geometry: 6 segments (four parallelogram segments, each with 67.5 degrees of
arc length and two trapezoidal key segments, each with 45 degrees of arc length).
– Radial and circumferential joint surfaces will utilize a compressional packing sheet to
enhance distribution of thrust between segments and minimize concentrated point loads.
– The 4-hour concrete strength is 2,000 psi, suitable for segment demolding and stacking.
Figure 1. Design Rendering of Precast Concrete Segment Rings for 3RPORT Tunnel.
1918
Table 1. Summary of load cases for TBM segment analysis.
Load Case Numbers Load Case Description
Production/transient stages
Load Cases 1-4 Segment stripping, storage, transportation, & handling
Construction stages
Load Case 5 TBM thrust jack forces
Load Case 6 Tail skin back-grouting pressure
Load Case 7 Localized back-grouting (secondary grouting) pressures
Services Stages
Load Case 8 Earth pressure, groundwater pressure, and surcharge loads
Load Case 9 Longitudinal joint bursting pressure
2.3 ACI Code Requirements
ACI 350 (2006) was designated as the concrete code for design of the precast concrete segments.
The code governs the design of liquid-retaining structures that “. . .are subject to different loadings,
more severe exposure conditions, and more restrictive serviceability requirements than non-envir-
onmental building structures.” Its objective is production of dense and durable concrete, suitable
for the exposure conditions, limiting deflections and controlling crack width through requirements
for concrete composition and spacing, size and distribution of reinforcement. However, it does not
contain requirements for elements using steel fiber reinforcement.
2.4 ACI 544.7R-16 Guidelines
ACI 544.7R-16 (2016) – Report on Design and Construction of Fiber-Reinforced Concrete
Tunnel Segments, was used to address requirements for design of fiber-reinforced segments, as
was information from various European codes for concrete design using fibers. This ACI
report notes the increased use of fiber reinforcing due to its significantly improved post-crack-
ing behavior and crack control characteristics. These characteristics are ideal for producing a
robust product that meets the performance requirements of ACI 350 (2006), which is more
resistant to handling stresses and has better long-term durability than conventionally
reinforced concrete. The ACI 544 report provides a procedure for structural analysis and
design based on governing load cases; and a description of material parameters, test, and ana-
lyses required for design. The load cases used in the subject design are listed in Table 1.
ACI 544.7R-16 also sets the framework for determination of concrete properties to be used
in the analysis, as derived from flexural strength testing performed on cast beam samples in
accordance with the requirements of ASTM C1609. This test determines a load-deflection
curve showing peak and post-peak behavior for third-point loading on a simply-supported
beam over a defined range of deflections. Typical beams used in the test are 6 inches square
and 18 inches long, and test loads at peak strength and at deflections of L/600 (0.75 mm) and
L/150 (3.0 mm) are reported. These test loads are used to calculate the first peak flexural
strength, f1 and the residual flexural strengths at deflections of L/600 and L/150, e.g., f D150.
ACI 544.7R-16 (2016) notes that in elastic analysis, the key parameter is residual tensile
strength, σp, which is calculated from flexural strength using an adjustment factor of 0.33 to 0.37.
The nominal resistance bending moment, Mn, is calculated using either Mn = σp bh2/2 or MN =
f D150 bh2/6. ACI 544 also gives recommended load factors for the various load cases and strength
reduction factors (ϕ) for various behavior modes (flexure, compression, shear, and bearing).
3 DESIGN APPROACH
3.1 Overall design approach
The design was organized to address the ACI 544.7R-16 (2016) load cases listed above. The
load cases are grouped in Table 1 by stages from production through construction, to service.
1919
The segments were designed to meet the required load factors of ACI 350 (2006), some of
which are higher than those of ACI 544.7R-16. The analysis also proceeded in a different
sequence than indicated by the case number in order to define the segment geometry based on
the most significant loadings.
3.2 Load Case 8 – Earth, groundwater, and surcharge loadings
This load case includes loadings on the tunnel lining due to the pressure of overlying rock and
soil, groundwater, and surcharge loads on the ground surface. The external loads were devel-
oped using the finite element (FE) code PLAXIS. The tunnel profile was reviewed in context
with the GBR and selected borings and test data to identify critical sections for analysis and
to develop overburden and rock mass properties for use in the PLAXIS modeling.
ACI 544.7R-16 (2016) load factors are lower than the 1.6 required by ACI-350 (2006) for
earth and groundwater loadings. Accordingly, the PLAXIS output loadings were factored by
1.6 and the results were evaluated without attempting to separate the effects of the various
components of the loads. This was performed using a bending moment-axial force (M-N)
interaction diagram for fiber-reinforced concrete, constructed in accordance with procedures
described in Appendix A of ACI 544.7R-16.
3.3 Load Cases 6 and 7 – Grouting Loads
Case 6 determines the concentric loads from the annular tail void grouting process.
Because the baselined groundwater pressure is 6.5 bars, and the anticipated grouting pressures
are up to 2 bars above that, a grouting pressure of 8.5 bars was used for analysis. Because the
segmental ring is surrounded by semi-liquid and fresh grout during application of grouting pres-
sure, no interaction is considered between the segmental ring and the ground. Radial pressure is
applied linearly from minimum grout pressure at the crown to maximum at the invert, with only
the lining self-weight of the lining and the grout pressure considered as shown in Figure 2a.
Load Case 7 evaluates the effects of localized secondary contact grouting. This grouting is per-
formed to verify closure of the annular void and is modeled by applying the differential grout pres-
sure of 2 bars over 10 percent of the segment ring at the tunnel crown, as shown in Figure 2b. The
tunnel lining is considered to be in full contact with the surrounding ground except in the area
where the secondary grouting is performed. The interaction between the lining and the surround-
ing ground is modeled using radial springs. The results were checked using the higher load factor
of 1.6 per ACI 350 (2006), using an M-N interaction diagram for FRC.
3.4 Load Cases 1-4 - Production, Handling, and Transportation Loads
Load cases 1-4 are evaluated using a spreadsheet for calculating bending moments from
simple segment and handling system geometry. Case 1 involves segment demolding, gripping
using cast-in pockets in the segment faces, and rotating the segment from the casting mold to
stacking and storage. Case 2 involves stacking the six segments of one ring using timber block-
ing between the segments. These cases are evaluated based on concrete strength at four hours.
Figure 2. Model for Cases 6 and 7 – annular backfill grouting and secondary contact grouting loads.
1920
Cases 3 and 4 are for transportation of the segments from the storage yard into the tunnel and
handling of the segments for placement during tunneling. These analyses utilize the flexural
strength parameters from concrete tested at 28 days, and they incorporate an impact factor of 2.0
in addition to the normal load factor to account for the nature of the construction process.
Code checks were also performed for shear around the cast-in segment rotator pockets, and
tensile pullout and shear around a center-pin lifting insert. Cases 1-4 and the related checks all
showed adequate capacity in the segments as designed.
3.5 Load Case 5 – Circumferential Joint Bursting Stresses
Load Case 5 evaluates the effects of the TBM thrust jacks against the most recently completed
ring of segments. High compressive stresses develop under the jacking pads and give rise to
significant bursting tensile stresses within the segment. Spalling tensile forces also act between
adjacent jacking pads along the circumferential joints.
ACI 544.7R-16 (2016) presents simplified equations for initial analysis of these forces based
on the ACI 318 (2014) concrete code and the DAUB segment guideline published by the
German Tunneling committee (2013). The ACI code is based on consideration of bearing on
concrete by the anchorage of a post-tensioning strand using a strut and tie model.
TBM thrust forces were estimated considering the weight of the TBM and back-up, the
baseline groundwater pressure of 6.5 bars at the tunnel face, the maximum excavation force
for each cutter, and the number of cutters. Frictional loads on the tunnel shield were estimated
using a coefficient of friction of 0.15 for normal conditions and 0.40 for upset conditions. A
load factor of 1.2 was used for normal friction per ACI 544.7R-16 (2016) and 1.0 was used for
upset friction. The normal condition with 1.2 load factor resulted in a maximum thrust of
34,493 kN, which is similar to the maximum thrust capacity of the TBM. Although ACI 350
(2006) requires a live load factor of 1.6 because these loads may have significant uncertainty,
this was not considered applicable because the maximum TBM thrust load was considered to
be more predictable, and a load exceeding that is considered unlikely. A capacity reduction
factor of 0.65 was used per ACI 350 because the action of the load for this case is compressive
in nature.
The simplified equations of ACI 544.7R-16 (2016) were evaluated using spreadsheets. The
results of these simplified analyses showed a high demand for strength required to resist burst-
ing stresses, which was not satisfied by the residual tensile strength of the FRC segments.
Therefore, based on the simplified analyses, conventional steel rebar reinforcement would be
required in addition to fiber reinforcements.
In order to further explore whether rebar was actually needed, a FE analysis was per-
formed using the commercial software Straus7 (2015), distributed by Strand 7 Pty Ltd,
Sydney, Australia. The intent of the FE analysis was to examine more closely the concrete
stress-strain behavior and extent of possible cracking that may occur. This included develop-
ment of a constitutive model based on flexural beam testing information to represent the
entire range of compression and tension behavior of the FRC segments. The constitutive
law incorporates the capacity reduction factor ϕ = 0.65 because of the compressional nature
of the loading.
3.6 Load Case 9 – Longitudinal Joint Bursting Forces
Load Case 9 addresses bursting forces in the longitudinal joints due to service earth and ground-
water loadings. The methodology for the assessment is essentially the same as for Load Case 5.
4 DEVELOPMENT OF CONSTITUTIVE MODEL
The stress-strain relationship for the FE analysis was chosen to simulate the expected behavior
of the FRC before and after flexural yield. The constitutive law aims to reproduce actual
stress-strain response of uniaxial tests; however, experimental data are usually obtained from
1921
flexural tests due to the difficulty of gaining significant results from uniaxial tensile tests. Con-
sequently, the “flexural tensile” parameters need to be converted into “equivalent tensile”
parameters.
A literature review indicated that one of the most widely-used approaches is that of
RILEM Technical Committee 162, Test and Design Methods for Steel Fiber-Reinforced
Concrete (2003). In this approach, tensile stress parameters are derived from flexural test
results, after scaling by proper stress coefficients. Relationships connecting tensile stress
values with flexural tensile values vary depending on the type of test beam used in different
test practices.
ACI 544.7R (2016) suggests scaling the post-crack (residual) flexural strength parameters
by an adjustment factor to convert the residual flexural parameters from the standard flexural
test to residual tensile strength for design. This scale factor is used to account for the nonlinear
stress distribution in the critical section and is only applied to the residual strengths. The first
peak flexural strength is not scaled because it occurs before the onset of cracking and a linear
stress distribution is expected at this stage.
RILEM TC 162 (2003), proposes a stress coefficient of 0.45 for the first residual strength at
CMOD1 (crack mouth opening displacement =0.5mm), which approximately corresponds to
the residual strength at L/600 according to ASTM C1609 (2012). RILEM proposes a coeffi-
cient equal to 0.37 for the residual strength at CMOD4 (crack mouth opening displacement
=3.5mm), which approximately corresponds to the residual strength at L/150 according to
ASTM C1609.
The tensile branch of the design stress-strain is determined as follows:
– The first peak design strength is obtained by multiplying the specified value by the concrete
strength factor φ;
– The residual design strength at L/600 is obtained by multiplying the specified value by the
concrete strength factor φ and the scaling factor 0.45;
– The residual design strength at L/150 is obtained by multiplying the specified value by the
concrete strength factor φ and the scaling factor 0.37.
A parabolic pattern is assumed for the compressive stress-strain relationship until a deform-
ation of 0.2%, after which the stress remains constant. The maximum compressive strain is
taken 0.3%, instead of the 0.35% value suggested by RILEM (2003), in order to respect the
convention according to ACI 318 (2014). The derived constitutive law is shown in Figure 3.
This constitutive law corresponds to an ultimate limit state (ULS) because it has been
reduced by the partial material factor for concrete.
Figure 3. Derived constitutive relationship.
1922
5 NUMERICAL (FINITE ELEMENT) ANALYSIS
5.1 Considerations for finite element analysis
FE analysis offers several advantages compared to the simplified equations. In addition to the
bursting tensile stresses, it can determine the spalling stresses expected in the areas between
the jacking pads. It can follow the post-crack behavior of FRC if the first peak strength is
exceeded, which allows evaluation of stress redistribution capacity due to stable crack propa-
gation. Simplified equations, by contrast, only compare the maximum elastic stress with the
FRC tensile strength to evaluate if conventional reinforcement is required. Finally, FE ana-
lysis enables obtaining more precise stress propagation because the actual segment and jack
shoe geometry, and the eventual load eccentricity, are all included in the model.
5.2 Model Details
Modeling was performed for the larger rhomboidal tunnel segment, which is contacted by
three, equally-spaced jack shoes. The segment geometry was simplified for modeling by
straightening the angled sides out; this simplification does not affect the results. The model
mesh size is approximately 30 mm, resulting in a total of about 50,000 brick elements. Stress
relief recesses near the segment intrados and extrados were modeled by removing correspond-
ing brick elements to facilitate stress distribution from TBM jack pressure deep into the seg-
ment to avoid minor concrete edge spalling that was not the focus of this analysis.
For the restraints, the longitudinal degree of freedom of all segment joints on the rear seg-
ment face in contact with the previously installed ring was fixed. The segment is conservatively
left free to move in the circumferential direction, whereas in reality it is partially restrained by
adjacent segments. This will result in more conservatively modeling bursting and spalling
stress effects and actual behavior should be better than estimated from the numerical analysis.
The TBM force was applied as equivalent normal pressure on the jack shoe. The factored force
per shoe is 2155.8kN. After division by the shoe area, a contact pressure approximately equal to
20 MPa is found, as shown in Figure 4, which also shows the simplified segment geometry. The
total design force on the segment is about 6467kN. This design force is not applied all at once, in
order to ease convergence, but rather over 20 simulation steps with a 5% load increment.
5.3 Model results – Bursting stresses
Initial results were encouraging, with the pattern of stresses associated with application of
60% of the design pressure shown in Figure 5, the pattern strongly resembling that of the
example analysis described in ACI 544.7R-16 (2016) for this load case.
Figure 4. FE model of segment with TBM shoe forces applies.
1923
Figure 5. Model of segment showing stresses 60% load application.
Figure 6. Graph of bursting forces from the FE FRC segment model.
Figure 6 shows that estimated bursting stresses are well below the first peak strength for the
portion of analysis past the first peak load which occurs between load Steps 13 and 14, and
for the portion past 100% of the TBM thrust load. Thus, they do not require evaluation
against residual flexural or tensile strength values. This also shows that the results from the
more precise FE model are much less than those estimated using the simplified analyses, the
results of which indicated that the bursting strengths were exceeded and rebar reinforcing was
necessary.
5.4 Spalling tensile stresses
Besides bursting stresses, spalling tensile stresses are also considered in the analysis. Contours
of tangential stress and strain at the next load simulation step, the 65% load step, show the
beginning of crack formation in the joint, at the midpoint of the spaces between the jacks,
exactly where spalling tensile stresses are predicted, as shown in Figure 7.
This is further illustrated in Figure 8 showing how the tangential stress develops in the spall-
ing area during the TBM thrust load application. The spalling stress increases until Step 13,
the 60% load step, where the peak stress defined in the constitutive law is reached. In the
1924
Figure 7. FE Segment model showing the development of crack formation between TBM shoes.
Figure 8. Tangential stresses in spalling area during TBM thrust load application.
following step, which marks the onset of cracking, a sharp decrease in stress occurs and there
is a corresponding strain increase. Then, until the total TBM thrust is reached the element
stress falls off slightly at a magnitude near the residual tensile strength.
Using this information, the maximum expected crack width at the full TBM thrust load was
estimated and compared with an acceptability limit. The analysis considers a single crack,
which in reality could be several smaller cracks in the area rather than one at the midpoint.
The crack affected a single element width, so the crack width is estimated by multiplying the
maximum strain by the element size:
Wmax ¼ 30 mm x 2:22E 03 ¼ 0:07 mm ð1Þ
It should be noted that this analysis, as an ULS analysis, would not normally be used for
studying crack width, which is normally considered at the service limit state (SLS). Both U.S.
and European codes limit crack width to about 0.3 mm for ordinary exposure conditions,
with values of about 0.10 mm for environmental water-retaining structures, for an SLS ana-
lysis. As the designed segments met this criterion for a ULS analysis, the segment was
1925
considered capable of withstanding the TBM thrust load. Two factors further mitigating the
estimated crack width are the likelihood that multiple, smaller cracks would actually occur,
and the confinement that would occur due to adjacent segments that was conservatively not
modeled.
6 CONCLUSIONS
Both simplified analyses and numerical FE analyses have demonstrated that the TBM thrust
force can be safely applied to the FRC segments. The simplified analyses were used to check
the majority of the load cases and found that the FRC was adequate. For load cases where
the simplified analyses indicated additional reinforcement may be necessary, the FE analyses
was used to further assess the adequacy of the FRC and demonstrate that rebar reinforcing
was not needed. The numerical model has been validated through comparison with analytical
solutions and analogue numerical results found in the literature.
The bursting stresses estimated in the FE analyses remain below the first peak strength, indi-
cating that under 100% of the TBM thrust these stresses are still in the elastic range. The most
critical aspect during TBM thrust consists of the spalling stresses between adjacent jack pads.
Until approximately 60% of the design thrust, the stresses are within the first peak limit. Above
this load value, the peak strength is reached and very small cracks (less than 0.1mm) are
expected to develop in the spalling area; however, the crack width is far below the acceptability
limit given in both European and American Standards. The crack width is also a maximum
value, estimated for one crack, whereas the actual condition may be multiple smaller cracks.
Because the predicted maximum crack width under ULS conditions is well below the allow-
able crack width, the use of conventional steel rebar reinforcement is not considered necessary.
Therefore the use of FE modeling has demonstrated that the FRC segment of 304.8mm thick-
ness is sufficient to withstand the expected loads without need of adding conventional
reinforcement.
REFERENCES
American Concrete Institute (ACI). 2016. Report on Design and Construction of Fiber-Reinforced Precast
Concrete Tunnel Segments. ACI 544R-16. Farmington Hills, MI
American Concrete Institute (ACI). 2014. Building Code Requirements for Structural Concrete. ACI 318-
14. Farmington Hills, MI
American Concrete Institute (ACI). 2006. Code Requirements for Environmental Engineering Concrete
Structures. ACI 350-06. Farmington Hills, MI
ASTM International (ASTM). 2012. C1609 Standard Test Method for Flexural Performance of Fiber-
Reinforced Concrete (Using Beam with Third-Point Loading). West Conshohocken, PA.
Deutscher Ausschuss fur unterirdisches Bauen e. V. (DAUB) German Tunneling Committee (ITA-
AITES). 2013. Recommendations for the Design, Production, and Installation of Segmental Rings.
Cologne, Germany.
Reunion Internationale des Laboratoires et Experts des Materiaux (RILEM), Technical Committee 162.
2003. Test and Design Methods for Steel Fiber-Reinforced Concrete. In Materials and Structures Vol.
36: 560–567. Paris, France.
1926
